WobblyRaccoon
WobblyRaccoon

Do you agree with this?

Anyone experienced with Deeptech startups, how is the quality of work and development?

Post image
9mo ago
Find out if you are being paid fairly.Download Grapevine

If I had a dollar for every noogler out there trying to start their own "AI" startup then I would have enough funding to start my own AI startup

GoofyDonut
GoofyDonut

"AI Startup"

unmasks

Boilerplate front-end with API calls to openai

CosmicLlama
CosmicLlama

No comment on the AI thing... But generally speaking we Indians lack imaginative entrepreneurial skills. Nothing wrong with that.. US acts as the lab and innovates new idea/business, we get to pick best model and make money. Untill that lab closes, a new lab will not open.

BouncyMarshmallow
BouncyMarshmallow

Case in point, the joke of llm - KRUTIM, sala api call laga diya, wrapper mey pack karkey dhol peet diya , nasNULList llm🤣

PrancingPotato
PrancingPotato

API call? Care to elaborate please.

BouncyNarwhal
BouncyNarwhal

API calls to OpenAI

MagicalQuokka
MagicalQuokka

I think it's too late. Need many hundreds of millions in investment to acquire hardware, data and talent. It can be done if someone is willing to put in the bucks and get the right talent.

PrancingPotato
PrancingPotato

Agreed partially. Seems like a lot of the technologies that will see significant advancements in the next 10-20 years require high capex. Which maybe another reason why Indian startups seem to be wrappers or cut paste ideas that require lesser capital.

MagicalQuokka
MagicalQuokka

Yup. You just cannot build foundation models without atleast many tens of millions in seed funding. Indian seed-stage capital is not reaching that ticket size soon. Purchasing power parity is against our interests here.

But if someone were to invest hundreds of millions into hardware (like Yotta Data, Hiranandani did), they might get get 10x of that in returns in the next 5 years.

SparklyBoba
SparklyBoba

i agree somewhat, AI do has increased my productivity but still output given by AI is unreliable. And if you want to increase the efficiency and all, you have to buy better gpu and other technical stuff which as of now isn’t value for money. Let see how does it improve

QuirkySushi
QuirkySushi

Isn’t this obvious?

Where are the datasets, the chips, the researchers, the capital?

Indian government is still figuring out who will ping the fans to build some chips while in the Silicon Valley you’ll find there are at least 10 startups that are pushing the envelope on edge, cloud inference speeds. Example — hailo.ai

By the time we figure out the hardware, the world would have moved doing next gen of models.

There are great Indian startups that are building AI tools but that’s what they are — tools to work on the application later. No foundation model. No full stack hardware solutions, and obviously no new research.

Discover more
Curated from across